乙型肝炎肝纤维化列线图预测模型的构建与评估  

Establishment and validation of a predictive Nomogram for liver fibrosis in patients with hepatitis B

作　　者：汤影子 郭艳 段海珊 何小勤 夏杰 TANG Yingzi;GUO Yan;DUAN Haishan;HE Xiaoqin;XIA Jie(Department of Infectious Diseases,the First Affiliated Hospital of Army Medical University,Chongqing Key Laboratory of Viral Infectious Diseases,Chongqing 400038,China)

机构地区：[1]陆军军医大学第一附属医院感染病科,病毒传染病精准防治重庆市重点实验室,重庆400038

出　　处：《胃肠病学和肝病学杂志》2025年第3期412-417,共6页Chinese Journal of Gastroenterology and Hepatology

基　　金：十三五传染病国家科技重大专项(2018ZX10723203-004-004)。

摘　　要：目的建立乙型肝炎肝纤维化列线图预测模型并验证其诊断效能。方法收集2010年1月至2022年6月在我院进行肝活检的1032例乙型肝炎患者的临床资料,随机分为训练组和验证组。在训练组中通过单因素分析及LASSO回归提取出有统计学意义的变量,进一步通过多因素Logistic回归确定乙型肝炎患者肝纤维化的独立危险因素,并构建列线图预测模型。采用ROC曲线和校准曲线评估预测模型的诊断效能,采用决策曲线分析列线图模型的临床实用性;采用ROC曲线、净重新分类指数(net reclassification index,NRI)及综合判别改善指数(integrated discrimination improvement,IDI)比较新建立的联合模型与APRI、FIB-4、TE之间预测能力的准确性。结果多因素Logistic回归分析提示,TE、血小板计数、白蛋白/球蛋白、凝血酶原活动度、甲胎蛋白、门静脉流速、脾脏厚径是乙型肝炎患者发生肝纤维化的独立危险因素(P<0.05)。列线图模型在训练组及验证组人群的ROC曲线下面积分别为0.886、0.866,同时该模型在训练组和验证组中均具有良好的校准度和临床实用性;列线图模型比APRI、FIB-4及单独TE参数模型的预测能力均有所改善(P<0.001)。结论本研究建立起一个相对准确的乙型肝炎肝纤维化预测模型,血清及影像学指标联合TE可提升预测模型的诊断效能。Objective To establish and validate predictive Nomogram for liver fibrosis in patients with hepatitis B.Methods The clinical data and laboratory indicators of 1032 patients with hepatitis B who received liver biopsy in our hospital from Jan.2010 to Jun.2022 were collected,and the patients were randomly divided into training group and validation group.Statistically significant variables were extracted from the training group through univariate analysis and LASSO regression,and independent risk factors for liver fibrosis in hepatitis B patients were further screened by multivariate Logistic regression,and a Nomogram prediction model was established.ROC curve and calibration curve were used to evaluate the accuracy and stability of Nomogram in the training group and validation group,and decision curve was used to analyze the clinical practicability of the Nomogram model.ROC curve,net reclassification index(NRI)and integrated discrimination improvement(IDI)were used to compare the predictability of the newly established combined model with that of APRI,FIB-4 and TE.Results Multivariate Logistic regression suggested that TE,platelet count,albumin-globulin ratio,prothrombin activity,α-fetoprotein,portal vein flow velocity and spleen thickness were independent risk factors for liver fibrosis in hepatitis B patients(P<0.05).The area under ROC curve of the Nomogram model in the training group and the validation group was 0.886 and 0.866,respectively.The predictive Nomogram showed good calibration and clinical utility in both the training group and validation group.Compared with APRI,FIB-4 and TE parameter model,the Nomogram model showed better predictive ability(P<0.001).Conclusion This study has established a relatively accurate predictive model for hepatitis B-related liver fibrosis.The combination of serum and imaging indexes with TE can improve the diagnostic efficiency of the predictive model.

关 键 词：乙型肝炎 肝纤维化 列线图 预测模型 

分 类 号：R512.6[医药卫生—内科学] R575.2[医药卫生—临床医学]